local determinants of economic structure: evidence from ... · revenue structure and the time...
TRANSCRIPT
Local Determinants of Economic Structure:Evidence from Land Quota Allocation in China
Meina Cai∗
Department of Political ScienceUniversity of Wisconsin-Madison
Email: [email protected]
November 2011
Abstract
China’s economic structure stands in stark contrast to transitional countries and manydeveloping countries at similar levels of economic development in its emphasis on industry andrelative underdevelopment of the service sector. This distorted structure indicates that economicreforms require a shift in resource allocation among sectors of the economy. This paper focuseson land, a fundamental means of production, which remains under state and collective ownershipmore than three decades after economic reforms were launched. In the face of massive arableland loss, the central government imposed a land quota system that restricts the maximumamount of land used at the subnational level, but gives local governments enough autonomy todetermine sectoral allocation of land quotas within their jurisdictions. What political economicfactors drive sectoral allocation of land quotas in China? This paper argues that both the localrevenue structure and the time horizon of local politicians have impacts on local land quotaallocation. Using an original dataset for a probability sample of 120 cities, this paper finds thatmore land quotas are distributed to industry when the local revenue base relies more on thevalue-added tax and less on business tax, when local Party secretaries, but not mayors, havelong time horizons, and when the locality is assigned more quotas.
∗The author is grateful to Melanie Manion, Scott Gehlbach, David Weimer, Yang Yao, Jean Oi for helpfuldiscussions. An earlier draft was presented at the Center for Chinese Studies 50th Anniversary Conference in honorof Kenneth G. Lieberthal, University of Michigan, Ann Arbor, Michigan, October 2011. Financial support fromthe National Science Foundation, the Chiang Ching-Kuo Foundation, the Institute for Humane Studies, and theUniversity of Wisconsin-Madison is gratefully acknowledged. All errors remain my own.
1 Introduction
The structure of centrally planned economies was distorted compared to market economies,
primarily reflected by a relative overdevelopment of heavy industry and underdevelopment of
services (Ofer 1987; Kornai 1992; Roland 2000). The transition from a command to market economy
thus requires a shift in resource allocation among sectors of the economy. The so-called Washington
consensus, with strong emphasis on liberalization, stabilization, and privatization, was carried out in
the process of transition in many communist countries in the former Soviet Union and East Europe,
with the expectation that this reform would induce the seriously distorted economic structure to
move in the direction consistent with market economies. Deviating from this standard reform
recipe, China developed some unique institutions, such as dual-track liberalization and township
and village enterprises (TVEs), while still maintaining its authoritarian control by the Chinese
Communist Party (CCP). Nonetheless, China has impressed the world by its astonishing economic
success, with an average growth rate of about 10 percent for the past 30 years. In 2010, China
surpassed Japan to become the world’s second largest economy behind the United States. China
is therefore often viewed as a great success, at least from the perspective of economic transition.
Contrary to expectations, after more than three decades of economic reforms, the Chinese
economic structure still stands in stark contrast to transitional countries as well as many developing
countries at similar levels of economic development. While postcommunist countries experienced a
significant structural shift from industry to services, China still maintains the bias toward industry
and relatively small share of services: between 1990 to 2009, while value added by industry as a
share of GDP declined from 46 to 29 percent in East Europe, from 48 to 33 percent in Russia, and
from 39 to 35 percent in other countries in the former Soviet Union,1 in China it increased from 41
to 46 percent, as shown in Figure 1. Within the same period, value added by services as a share
of GDP grew much faster in postcommunist countries than in China: the ratio increased from 39
to 64 percent in East Europe, from 35 to 62 percent in Russia, and from 32 to only 43 percent in
China, as shown in Figure 2. Holding time invariant, China’s economic structure still stands out
in comparison with developing countries at similar level of economic development, measured by
1The countries in East Europe and the Baltics in the dataset include Albania, Bulgaria, Croatia, Czech Republic,Estonia, Hungary, Latvia, Lithuania, Macedonia, Poland, Romania, Slovenia, and Slovakia. The countries in theCIS in the dataset include Azerbaijan, Armenia, Belarus, Georgia, Kazakhstan, Moldova, Tajikistan, Turkmenistan,Ukraine, and Uzbekistan.
1
GDP per capita in 2005 purchasing power parity (PPP) dollars (Figures 3 and 4). Those showing
a similar pattern of economic structure with China are often oil-rich countries, such as Angola
and Azerbaijan. Both cross-time and cross-sectional comparisons suggest that China’s economic
structure is far from optimal. Huang and Tao (2010) found that economic structure in China suffers
from serious distortions in labor, capital, land, energy, and the environment.
[Figures 1-4 about here]
This paper examines the political economic determinants of land allocation between industry
and services in China. The focus is on land because it is one of the fundamental means of
production allowing economic activities to take place. More importantly, land remains under state
and collective ownership and thus is an instrument with which local governments intervene in the
economy. With the emergence of a land market where land use rights can be leased out for a
period of time varying from 40 to 70 years, land has become an important revenue source for local
governments. In 2009, the land conveyance fee alone – a form of revenue generated by transferring
land use rights from the state to other land users – was up to 1424 billion RMB, accounting for 77.7
percent of the state fund budget,2 or equivalent to 20.8 percent of the state budget.3 Land also
functions as collateral to help local governments apply for bank loans to develop local infrastructure
(Whiting 2011).
China’s arable land loss parallels its economic growth, as shown in Figure 5. In the face
of massive arable land loss, the central government imposed a land quota system that restricts
the maximum amount of land used at the subnational level, but gives local governments enough
autonomy to determine sectoral allocation of land quotas within their jurisdictions. What political
economic factors drive sectoral allocation of land quotas in China? This paper argues that both the
local revenue structure and the time horizon of local politicians have impacts on local land quota
allocation between industry and service sectors. Using an original dataset compiled from various
statistical yearbooks and career histories of local politicians for a probability sample of 120 cities,
this paper finds that more land quotas are distributed to industry when the local revenue base
2Fund budget or jijin yusuan is another important revenue component for the state, in addition to the budgetrevenue. Fund budget is collected from the society through land use rights transfer and public lotteries. It is used tosupport specified infrastructure and social development.
3Ministry of Finance, available at www.mof.gov.cn/zhengwuxinxi/caizhengshuju/201005/t20100511 291390.html
2
relies more on the value-added tax and less on business tax, when local Communist Party leaders
have long time horizons, and when the locality is assigned more quotas.
[Figure 5 about here]
The paper ties in with three strands of literature. A first strand builds on the grabbing hand
model of government by Andrei Shleifer (1997) and his various coauthors (Murphy, Shleifer and
Vishny 1991; Frye and Shleifer 1997; Shleifer and Vishny 1998). A general consensus in this
literature is that political factors matter enormously for resource allocation, driving economic
outcomes deviating from the Pareto optimum. For instance, Murphy, Shleifer and Vishny (1991)
examine the allocation of human capital and find that in countries where rent seekers can claim
a substantial part of wealth through official and unofficial expropriation, the most talented people
join rent-seeking rather than entrepreneurial activities, thereby reducing economic growth. Recent
experiences of transition economies show that the state is a central barrier to economic development
in postcommunist countries (Shleifer 1997; Frye and Shleifer 1997). Although Poland and Russia
adopted similar reform packages, Russia lags significantly behind Poland in the transition of its
government. Consequently, the Polish economy prospered while Russia continues to produce the
wrong products.4 Gehlbach (2008) provides a political economy logic of the distorted Russian
economy. He argues that politicians have an incentive to disproportionally provide collective goods
to sectors that are important sources of state revenue and are anticipated to be more tax complaint.
As a result, in the former Soviet Union, where the revenue base was largely inherited from the old
regime, politicians were systematically biased against the new private enterprises that are less
taxable, whereas in Eastern Europe, the revenue base was restructured to include the new private
sector, thereby encouraging politicians to promote the new economy. Along this line, this paper
examines the role that local governments play in allocating land quotas.
The second strand in the literature begins with the pioneering work of Mancur Olson (1993)
examining the incentive structure of politicians. The state is after all staffed by politicians who
are themselves subject to institutional restrictions affecting their time horizon. Olson’s analogy of
roving and stationary bandits best illustrates the logic of time horizon. A “roving bandit” with
4For instance, in Russia the vodka sector was created in localities where comparative advantage in vodka industryis absent (Gehlbach 2008).
3
a short time horizon has a strong incentive to confiscate wealth and abrogate contracts without
considering the long-run consequences of his choices. Such predation destroys the incentive for
citizens to engage in productive activity, contributing to a decline in economic output. In contrast,
a “stationary bandit” with a long time horizon has an incentive to provide public goods, respect
individual property and contract enforcement rights so as to extract the maximum possible net
surplus in the long run. This action encourages citizens to invest and produce, and consequently
generates economic growth under a dictatorship. Clague et al.(1996) find a compelling empirical
evidence that the time horizon of politicians has an impact on property and contract rights. Wright
(2008) demonstrates that autocrats with long time horizons use foreign aid more effectively than
those with short time horizons. This paper applies the logic of time horizons to the Chinese context
and finds support for this line of thought.
The third strand build on the vast literature on the Chinese political economy. China’s
economic success relies critically on local governments. The mechanism that aligns local government
incentives with promoting economic growth lies in the fiscal contracting system (Oi 1992; Montinola,
Qian and Weingast 1996; Oi 1999; Jin, Qian and Weingast 2005). This fiscal arrangement however,
was replaced by tax sharing system in 1994, which differs significantly from the previous fiscal
system. Moreover, previous literature examining how local governments promote economic growth
did not take land into account (Oi 1999; Whiting 2000); this is because at the time when this
research was conducted, the land market was largely absent: land was distributed to land users by
the state, free of charge. Although research on China’s land reform is now growing, it pays exclusive
attention to how land has become an important instrument for local government in promoting
the local economy (Lin 2009; Hsing 2010; Man and Hong 2011; Whiting 2011). The land quota
restriction has been largely ignored: a few studies have very briefly noted the land quota system
(Lin 2009; Hsing 2010), but failed to examine it in detail.
According to my interviews with officials across localities and across all levels of local
governments, land quotas appeared to be the largest constraint in promoting local economic growth,
indicating that land quotas have become a scarce resource to local governments.5 This is especially
5Interviews ZJ12160109, GD01311110, GD03040110, GD03040210, GD03050110, ZJ03190110, ZJ03190210,ZJ03290110, JS04020110, JS04060110, JS04071110, ZJ04120110, ZJ04130110, ZJ04160110, ZJ04160210, ZJ04190110,ZJ04190210, ZJ04200110, ZJ04200210, CQ04250110, CQ04280110, CQ05060110, CQ05060210, CQ05110110,CQ05121210. Interviews were conducted in Zhejiang, Jiangsu, Guangdong and Chongqing and cover a wide rangeof local leaders from provincial-level officials to village party secretaries. See appendix A for discussion of interview
4
the case along the east coast, where the economy has developed the most in China. For instance,
when I asked why a piece of unoccupied land located within an economic development zone could
not be used for industry in Wenzhou, the official responded: “We ran out of this year’s quotas,
and we are waiting for new quotas” (Interview ZJ04190210). My in-depth qualitative interviews
identify land quotas constraining local government officials in attempting to promote the local
economy. Using the grabbing hand model and the time horizon model as theoretical departures,
this paper examines how the new fiscal arrangement affects local politicians in sectoral allocation
of land quotas.
The contribution of this paper is three-fold. First, by highlighting the largely overlooked yet
important institutional constraint of land quotas, the paper contributes to the growing literature on
China’s land reform. Second, it systematically examines how the new fiscal system and land change
the dynamics of how the local economy is operated. By establishing empirically the relationship
between the fiscal system, land, and local resource allocation strategies, this paper updates the
literature on the Chinese political economy. Third, it offers a political economic logic of resource
allocation between economic sectors. It shows that the length of politician tenure is one of the
main driving forces, thus adding another layer to our understanding of the distortion of economic
structure.
The rest of the paper is organized as follows. Section 2 discusses local governments, the
main players considered in this paper, and their preferences. It examines the political, fiscal, and
land quota constraints facing local politicians. Section 3 details the political economy of sectoral
allocation of land quotas. It develops a simple two-period model to theorize the arguments and
derive hypotheses. Section 4 describes the data and construction of variables. Section 5 estimates
the effects of economic and political determinants on local land quota allocation and explores the
robustness of findings. Section 6 concludes.
2 Players, Preferences, and Institutional Context
This paper considers local politicians as the main players. A local politician is assumed to be a
revenue maximizer, subject to constraints. Social scientists have long stressed the importance of
subjects and interviewee coding.
5
revenue extraction. Edmund Burke (1790) succinctly asserted over two centuries ago:“The revenue
of the state is the state. In effect all depends upon it, whether for support or reformation.”6
Although it is common in the political economy literature to assume that a rational self-interested
ruler is a revenue maximizer (Bates 1981; North 1981; Levi 1988; Olson 1993), it still requires some
elaboration in the China context. Unlike democracies, where politicians are restricted by voters,
officials in authoritarian China are managed through the nomenklatura system, through which
the Communist Party holds the ultimate authority over personnel management (i.e., appointment,
promotion, transfer, or removal) of all Party and state main leaders (Manion 1985; Burns 1994;
Lam and Chan 1996). The career prospects of leading officials are based on their performance,
evaluated regularly using the target responsibility system (TRS), a set of performance criteria that
induce local officials to act in ways commensurate with the preferences of the Party-state (Tsui and
Wang 2004; Whiting 2000, 2004).
For example, the performance of local political leaders7 in a prefecture-level city in Guangdong
province is evaluated yearly based on the ten indicators detailed in Table 1. A target for each
indicator is set for each locality prior to the evaluation. Each indicator is assigned a baseline score:
a higher score means more weight in the overall score calculation. Score calculation varies across
indicators: some are dichotomous, indicating whether or not the target has been reached, while
others provide a relative measure, comparing different localities. An examination of the evaluation
scoring in Table 1 offers two stylized facts. First, local budgetary revenue is assigned the highest
score, indicating its priority status among the indicators. Recent statistical analysis shows that
provincial-level officials who contribute more revenue to the central government are more likely to be
promoted (Sheng 2011; Shih et al. 2011). Second, consistent with the literature (Edin 2003; Landry
2003, 2008; Li and Zhou 2005), the evaluation is biased toward economic performance: 5 out of 10
indicators are economy-related and the sum of the baseline scores of these indicators accounts for 50
percent of the total baseline score.8 The score calculation for the economy-related indicators takes
into account the ranking in target fulfillment across local jurisdictions, which promotes competition
6Quoted in Pollack 2009, p. 1.7To protect my source, I do not identify the city. The local political leaders subject to evaluation by the
city comprised the following positions from each district under its jurisdiction: party secretary and deputy partysecretary of a district’s party committee, members of the standing committee, director and deputy director of districtgovernment, people’s congress, people’s political consultive conference, and secretary and deputy secretary of thecommission for disciplinary inspection.
8I consider the first four indicators and the development of hi-tech zone to be economy-related.
6
across local jurisdictions. Evaluation results have an impact on individual wealth accumulation of
local political leaders and, more importantly, on their career advancement.
[Table 1 about here]
Despite being a top priority, the economy-related criteria are hard to fulfill due to substantial
budgetary revenue shortfall. The rapid economic growth in China since the 1990s has been
maintained by substantial urban bias, massive infrastructural investment, and preferential
treatment to large state-owned enterprises (SOEs) and foreign direct investment (FDI). This
form of growth is conceptualized as “state-led capitalism” that relies heavily on investments from
local governments (Huang 2008). The problem is that local politicians, who desperately need
revenue to promote the local economy, run budget deficits themselves. Economic competition
across jurisdictions motivates local politicians to provide preferential treatment (e.g., tax breaks)
to attract investors, resulting in lower local tax collection. More importantly, fiscal pressure
on local politicians was intensified by the tax sharing system (TSS) launched in 1994, which
fundamentally changed the dynamics of how central and local governments divide tax revenue.
The TSS successfully transferred revenue from local governments to the center, but it failed to
adjust expenditure responsibilities for local governments, as shown in Figure 6. As a result, local
politicians experienced a sharp decline in local revenue while they are still expected to take on the
same expenditure responsibilities for the provision of a wide range of public goods and services (e.g.,
education, healthcare, and social welfare), thereby creating the potential for substantial budget
deficits (World Bank 2002; Wong 2009). The deficit problem gets more severe as you look down
the administrative hierarchy, because a local government has to share its tax collection with all
local governments above it administratively, in addition to the central government (Oi and Zhao
2007).9 Figure 7 is a scatterplot of the budget deficit measured by the ratio of local expenditure to
budgetary revenue against local GDP in 2005 for a probability sample of 120 cities in China.10 Of
the 120 cities surveyed, 117 cities had ratios greater than 1, meaning they spent more than they
took in.11 The figure also indicates a negative correlation between the budget deficit and local
9There is some variation in how local governments at different levels share local revenue. For example, in Zhejiangprovince, local revenue at the county level is shared with the provincial government but not the municipal governmentadministratively above it.
10For further discussion on sampling selection of the 120 cities, see section 4.1 on data.11The three cities that ran budget surpluses are Hangzhou, Fuzhou, and Urumqi.
7
GDP: the richer the city, the smaller the fiscal gap between expenditure and revenue.
[Figures 6 and 7 about here]
The mismatch between subnational governments revenue and expenditure mandates leads
to greater reliance on intergovernmental fiscal transfers, which have not yet offset the local
fiscal gap (World Bank 2002; Whiting 2011). A large amount of intergovernmental transfers
are pegged to the size of the local government payroll, meant to guarantee the minimal
maintenance of local governments, leaving unsolved the problem of insufficient revenue to promote
the local economy (Shih et al. 2010). A combination of heavy expenditure responsibilities
and ineffective intergovernmental transfers creates revenue-starved local politicians who must
resort to extrabudgetary revenue to supplement inadequate government funding. A high-ranking
provincial-level official described his tough fiscal situation very bluntly: “The center takes the larger
piece of the local tax revenue pie, while the local government gets the smaller piece. However, we
local governments have to do all the work. We need money. We can rely on nothing but land, so
we grab land and we must maintain a monopoly over land” (Interview CQ05060110). Not only
does this quote describe the budget deficit problem facing local politicians, but it also provides the
single solution to the problem: land.
Local politicians are motivated to expropriate land, as reflected in the quote above. More
importantly, their land expropriation is protected by legal and state regulations. Under the existing
land tenure system, land is segmented into urban land owned by the state (meaning the local
government or some other government agency) and rural land owned by rural collectives. The state
can expropriate rural land for the sake of the “public interest” (Constitution, Article 10). Under
this principle, the state that claims to represent the “public interest” retains ultimate authority
over all land, urban or rural (Hsing 2010). Since the introduction of the land leasehold market
in 1988, land use rights are separated from land ownership rights and can be leased out in the
market for a fixed number of years.12 However, the sale of long-term leases for construction land
use rights is limited to the urban land market.13 The revenue from leasing urban construction land
12The land-lease time limit is determined by the purposes of land use. Land use rights can be leased out for 70years, 50 years, and 40 years when land is used for residential, industrial, and commercial purposes, respectively. SeeState Council, Interim Regulations on Transfers of Urban State-owned Land Use Rights, 1990, Article 12.
13From a land use perspective, land is categorized into agricultural land, construction land, and unused land. LandAdministration Law, Article 4.
8
use rights is received as a lump sum payment made at the time of the transaction, officially called
the “land conveyance fee” (tudi churangjin), and it constitutes the dominant component for local
extrabudgetary revenue. Governed by this dual land system, rural land must be first converted to
urban land in order to fully realize its market value. As a result, the state – the exclusive body with
the authority to expropriate land – obtains land from rural households and provides relatively low
compensation.14 The state, which monopolizes the primary urban land market,15 then leases land
use rights at a price substantively higher than the land compensation paid to rural households. The
price differential between rural and urban land arising from the distorted land market generates
monopoly rents, easily captured by local governments. In the past two decades, the difference
between the land compensation paid to farmers and the market price of the seized land is about 2
trillion RMB for 14.7 million hectares of land (China Daily 2010). Local politicians not only grab
land in the city outskirts by converting rural land into urban land, but also fight fiercely with other
urban land owners, or “socialist land masters,” who occupied their land through administrative
allocation in the pre-reform era (e.g., SOEs and military units) within urban areas (Hsing 2010).
Under the unique land tenure system, land has become a key source of fiscal revenue for local
governments (Perkins 2009; Man 2011; Whiting 2011). It helps local politicians relieve their fiscal
stress and fulfill unfunded expenditure mandates. Meanwhile, the rapid shrinkage of rural land has
become a major source of social instability and food insecurity, both of which concern the central
government. Forced eviction and inadequate compensation during land expropriation are the main
cause of social unrest. Reportedly, about 65 percent of mass protests in rural areas are triggered
by land disputes (China Daily 2010). The central government, concerned first and foremost about
regime stability, has aggressively increased its control over land supply.16 It announced a one-year
moratorium on arable land conversion to non-agricultural use in May 1997 and a six-month freeze
14Compensation is composed of land compensation fees, resettlement fees, and compensation for what was attachedon the expropriated land. The Land Administration Law specifies the compensation to rural households whose landis expropriated (Article 47).
15There are two types of urban land market: primary and secondary. The primary land market exists between thegovernment, the exclusive land use rights sellers, and land users who receive land use rights from the governmentthrough negotiation (xieyi), bid invitation (zhaobiao), auction (paimai), and quotation (paimai). In the secondaryland market, land use rights switch from one land user to another through transfer (zhuanrang), sublease (zhuanzu),and mortgage (diya).
16Deng Xiaoping emphasized the overriding importance of political stability. In the absence of a stable environment,China would not be able to achieve anything and might even lose what has been accomplished. See Yang 2004, pp.4-6.
9
on agricultural land conversion in April 2004.17
In 1998, the central government substantially revised the Land Administration Law to take
agricultural land preservation more seriously. The central government attempts to control the
total amount of construction land at the subnational level by setting mandatory land quotas.18
These quotas are distributed top-down, along the administrative hierarchy: the central government
sets national quotas and disaggregates them to provinces; each province then disaggregates its
quotas to its municipalities, and each municipality to its counties.19 To facilitate local compliance,
the central government has invested in satellite remote sensing technology to detect local land
violations. Land quota restrictions pressure local governments to use land more efficiently. Some
provinces (e.g., Zhejiang, Guangdong) have experimented with including land use efficiency as an
indicator in their performance evaluation of officials. Referring back to the evaluation scoring in
the prefecture-level city of Guangdong in Table 1, land use efficiency was a sub-indicator under
the category of development of a Hi-tech zone. Failure to meet the quota requirement results in
sanctions (e.g., warning) and a reduction in land quotas for the following year.20
To sum up, the main players in this analysis are politicians at the subnational level. Although
economic decentralization grants local politicians considerable discretionary power within their
jurisdictions, they are still subject to the political (i.e., control of career prospects), fiscal, and
land quota restrictions imposed by the central governments. Local politicians prefer to maximize
revenue not only because revenue is a crucial indicator in the cadre evaluation, but also because they
must first extract sufficiently large revenue to promote the local economy and meet other targets
specified in the evaluation so as to increase the likelihood of career advancement. As such, the
revenue-maximizing local politician assumed in this paper does not conflict with the office-seeking
17For the first moratorium on land conversion, see State Bureau of Land Administration and State PlanningCommission, Regulation on Freezing Non-agricultural Construction Project Occupying Arable Land, May 20, 1997.For the second freeze, see General Office of the State Council, Urgent Notification of Regulating Land Market andStrengthening Land Management, April 29, 2004.
18These land quotas include agricultural land conversion to construction land, arable land to be maintained,and arable land to be created through development and reclamation. See State Council, Regulations on theImplementation of the Land Administration Law, December 27, 1998, Article 13. These quotas are also specifiedin several local government documents, which are not publicly accessible, but are available in redacted form bycontacting the author.
19In practice, there is some variation in quota assignments across provinces. For instance, in Zhejiang, the provincialgovernment bypasses municipalities and directly assigns quotas to its counties (Interview ZJ04200110). The interviewsubject was a deputy mayor of a county-level city in Zhejiang.
20General Office of the State Council, Urgent Notification of Regulating Land Market and Strengthening LandManagement, April 29, 2004, Article 4.
10
local politician when he faces serious fiscal pressure.
3 Theory
3.1 Political Economy of Sectoral Allocation of Land Quotas
Revenue structure has an important impact on revenue extraction strategies adopted by politicians
(Easter 2002; Gehlbach 2008). A big challenge facing post-communist countries during the
transition to the market economy is to reconstruct the revenue base and restore capacity to raise
revenue, because economic reforms (e.g., liberalization and privatization) destroyed old revenue
sources under communism while creating new untapped sources (e.g., the private sector). In
China, similar to other communist countries, the tax base prior to economic reforms depended
overwhelmingly on a few thousand large state enterprises. In the late 1970s, over 80 percent
of budgetary revenues were extracted directly from the state industrial sector, in the form of
taxes and profits collected from state-owned enterprises (Naughton 1992). Unlike its communist
counterparts, however, the economy in China first gained its momentum in the rural areas through
the development of collectively-owned township and village enterprises (TVEs). Decollectivization
of agricultural production transferred the bulk of agricultural income from rural collectives to
individual households, forcing local cadres to develop alternative sources of revenue to supplement
the loss from the previous revenue base (i.e., revenue from agriculture) (Oi 1992, 1999). The
relaxation of state monopoly control of industry allowed non-state economic actors to enter into the
industrial sector, thereby providing a new revenue source for local governments (Naughton 1992).
Perhaps more fundamentally, the fiscal contracting system21 in 1978-1993 granted local governments
residual rights over the local revenue, thereby creating incentives for local governments to promote
the local economy (Oi 1992; Montinola, Qian and Weingast 1996; Oi 1999; Jin, Qian and Weingast
2005). Under this fiscal system, firms submitted their taxes and fees to the level of government
that owns them. TVEs thus became an important revenue source for rural governments. This
particular fiscal arrangement “allowed the local residual to grow to maximum proportions, even
at the cost of denying the central state maximum tax revenue. Localities were allowed to benefit
21The fiscal contracting system required local governments to submit only a portion of their revenues to theirsuperiors according to the contract with their superiors; local governments retained all, or at least most, of theremainder.
11
disproportionately from local economic growth. . . .The Chinese reforms succeeded in generating
local economic growth because the central state did not get the taxes right” (Oi 1999, 57). However,
not all local cadres chose to develop TVEs. In areas with a historical legacy of weak collective
enterprise development, most typically in Wenzhou in Zhejiang province, local cadres developed
their revenue sources by aggressively promoting the private sector (Whiting 2000).
Local revenue maximization did not translate into tax maximization, because local cadres shifted
their revenue source from tax to non-tax income by imposing various non-tax fees and levies, both
of which fall into the category of extrabudgetary revenue (Oi 1992). As a result, despite successful
rural industrialization, total government budgetary revenue nevertheless declined dramatically from
31 percent of GDP in 1978 to 11 percent in 1994 (China Fiscal Yearbook 2009, 475). In contrast,
extrabudgetary revenue grew from 10 percent of GDP in 1978 to 14 percent in 1992 (China Fiscal
Yearbook 2009, 497). Not only did budgetary revenue diminish, but the central government’s
share of budgetary revenue also declined dramatically. Beginning in 1994, when the tax system
was reformed, both budgetary revenue and the share claimed by the central government started
increasing.
Under the new fiscal scheme launched in 1994, local government’s tax base is comprised primarily
of business tax, value-added tax (VAT), and income tax from all enterprises other than central
state-owned enterprises. The three taxes combined account for 60-70 percent of local budgetary
revenue.22 Examining who pays what indicates that the major tax components vary considerably
across sectors.23 Both industry and service (tertiary) sectors pay corporate income tax, but the
former contributes to VAT, whereas the latter contributes to business tax with only a few exceptions:
the construction sector is categorized as industry but pays business tax, whereas the wholesale and
retail sectors are categorized as service sector but pay VAT. Table 2 provides details on tax variation
by sectors. Of all VAT contributors, manufacturing industry contributes the most: its contribution
to VAT was about 60 percent from 2001 to 2006 (China Tax Yearbooks 2002-2007).
[Table 2 about here]
22This range is calculated by the author from a dataset compiled from China Fiscal Yearbook, 1995-2009.23The economy is divided into three sectors: primary, industry, and tertiary (service). In China, the primary sector
refers to agriculture. The industry sector is comprised of mining, manufacturing, production and supply of electricity,gas, and water, and construction industries. All economic activities not included in the primary and industry sectorsfall into the category of the service sector, such as real estate, banking, and retail.
12
While local budgetary revenue is primarily composed of taxes, the land conveyance fee
dominates local extrabudgetary revenue. Similar to tax revenue, the land conveyance fee also
varies significantly across sectors. From a land use perspective, construction land is categorized
into land for industrial, commercial, and residential uses. Industry occupies land for industrial
use, whereas most sectors that fall into the category of tertiary industry occupy land used for
commercial and residential purposes. In general, the land conveyance fee generated from land for
commercial and residential use is much higher than that generated from land for industrial use, as
shown in Figure 9: the land profit generated by the former was four times the profit from land used
for industry in 2004 and almost eight times in 2008.
[Figure 9 about here]
Although the service sector generates land revenue that is more sizable and more immediate
than that generated by industry, such revenue is a one-time payment. Once the land transaction
is over, the land revenue dries up. The business tax generated by service sector has an initial
spurt but stabilizes at a low level afterwards (Tao et al. 2010). For instance, the real estate
sector, the primary business tax contributor, produces high business tax at the time houses are
sold. Unlike many developed countries, China has not introduced property tax. Once housing sales
are over, government business tax drops quickly. In contrast, industry usually cannot generate
high revenue at the early stage and often has a longer take-off period, but it can generate a steady
revenue stream in the long run. Moreover, the development of the manufacturing industry, the
primary VAT contributor, has a spillover effect on the economy. After manufacturing firms are
established, their employers create demand for housing, entertainment, shopping and so on, thus
contributing to the development of the local service sector. In short, there is a trade-off in the
local revenue contribution between industry and service sectors once time is taken into account:
the latter generates more land revenue, which is a one-time payment, but the former produces a
steadier revenue stream in the long run. Applying Olson’s logic to resource allocation in China now
becomes straightforward. A local politician with a long time horizon, like a stationary bandit, is
more likely to allocate more resources to industry that generates sustainable local economic growth
so as to profit from the larger revenue pie in the future. A local politician with a short time horizon,
like a roving bandit, will distribute more resources to the service sector that generates immediate
13
revenue.
In the Chinese context, there are two institutional arrangements of the personnel management
system that are expected to influence the time horizon of government officials: mandatory
retirement and rotation of officials. In an effort to streamline and rejuvenate the cadre corp,
age-based retirement from office was introduced to replace the de facto lifelong tenure in 1982
(Manion 1993).24 The rules generally set retirement ages at fifty-five for women and sixty for men.
Officials in specified positions of leadership retire later, at sixty or sixty-five, depending on position
(CCP Central Committee 1982, Article 3).25 Empirical survey data show that the implementation
of mandatory retirement policy improved over time and is now strictly enforced (Li and Zhou 2005;
Landry 2008).
Officials are constrained not only by age-based retirement that sets a maximum value on their
expected remaining political life, but also by the rule of rotation that sets a tenure limit for a
particular position. To prevent the entrenchment of local political bosses, a leading official of a
local party committee and government must be transferred if he or she has worked in the same
position for ten years. Officials can be transferred between different localities, departments, levels
of government, and across party, government, enterprises and public organizations (CCP Central
Committee 1995, Article 38).26 Positions in party committees and governments at the county level
or above have a term limit of five years (General Office of the CCP Central Committee 2006, Article
3 and 6). That is, no one can stay in the same position for more than his institutionally given two
terms. By the end of the second term in office, a politician faces one of the following political fates:
promotion within the same locality or elsewhere, transfer to a position of identical bureaucratic
rank, exit from office if he reaches retirement age, and, in the worst cases, removal from office for
24The central government started promoting retirement reform for officials in 1978. Between 1978 and 1981, theretirement age was set in regulations, but the criteria of poor health and inability to continue work, not old ageper se, were the actual determinants of retirement. In 1982 age-based retirement was established as a general rule,without reference to health or ability to perform official duties. For more discussion on age-based retirement, seeManion 1993.
25The following officials, men and women, retire at sixty-five: government ministers and Central Committeedepartment heads, provincial party committee first secretaries, and provincial governors. The following officials retireat sixty: deputy ministers and Central Committee department depute heads, provincial party committee secretaries(other than first secretaries), provincial deputy governors, bureau chiefs and their deputies in State Council andCentral Committee bureaus, heads and deputy heads in provincial party committee and provincial governmentdepartments, prefectural party committee secretaries and deputy secretaries, and prefectural mayors and deputymayors. See CCP Central Committee, Decision on Institutionalization of Retirement of Aging Cadres, February 20,1982, Article 3.
26The rotation requirement for officials was reiterated in 2002. See CCP Central Committee, Regulation on Selectionand Appointment of Leading Party and Government Officials. July 9 2002, Article 52.
14
some reason (e.g., corruption). It is expected that a local politician at the start of her term has
a long time horizon. As she moves to the end of her term, her time horizon shortens. Knowing
she will not serve in the same position after her term is over, she is less interested in maximizing
revenue in the long run.
In sum, the 1994 tax sharing system introduced significant changes to local revenue structure. It
not only redefined the fiscal relationship between the central and local governments, but also shifted
local revenue structure from an ownership-based structure to a sectoral-based structure. Local
politicians are thus expected to develop sectoral, rather than ownership, preferences in resource
allocation so as to generate revenue.
3.2 Model
A local politician is a revenue maximizer, subject to the land quota assigned from above. Let k
be the land quota available to her. The politician values all sources of revenue, budgetary and
extrabudgetary. Budgetary funds come from tax revenue collected from firms. Let τ be the tax
rate imposed by the higher-level government. Extrabudgetary funds are primarily derived from
leasing land use rights. Let s be the land conveyance fee collected per unit of land by the local
politician. Let c be the cost incurred by the local politician in preparing expropriated land for
leasing out. This cost includes compensation to previous land users as well as infrastructure built
on the land to sell. I consider two sectors in the economy: manufacturing industry (denoted as 1)
and tertiary industry (denoted as 2). A local politician assigns k units of land to a sector. I denote
the sector’s production function with k units of land input by f(k). I further define f1(k1) = A1k1
and f2(k2) = A2kα2 where 0 < α < 1. The two sectors take different functional forms in their
production functions because a manufacturing firm is unlikely to be sensitive to location. Hence,
a linear function is sufficient to capture the characteristic that every additional unit of land used
for industry produces a constant return. By contrast, land used for residential and commercial
purposes is sensitive to location: land closer to the city core is likely to be more expensive than
land further away from the city core, and this is captured by a concave function.
A local politician collects both tax (budgetary) and non-tax (extrabudgetary) revenues.
Consider a simple two-period model: in the first period, a local politician sells construction land
use rights and receives a land conveyance fee, but she cannot receive tax revenue because it takes
15
time for land users to generate tax. For instance, a land developer can generate tax revenue only
after she builds houses and successfully sells them. In the second period, the politician collects tax
revenue from both sectors, but not a land conveyance fee, which is a one-time lump sum payment
at the time of land transaction. A politician’s time horizon is captured by the discount rate d.
A local politician maximizes:
maxk1,k2
s1k1 + s2k2 − c(k1 + k2) +1
1 + d[τ1f1(k1) + τ2f2(k2)]
s.t. k1 + k2 ≤ k
The interior solution is as follows:
k∗2 =
(τ1A1 + (1 + d)(s1 − s2)
τ2A2α
) 1α−1
k∗1 = k −(τ1A1 + (1 + d)(s1 − s2)
τ2A2α
) 1α−1
(1)∂k∗1
∂(τ1A1)> 0,
∂k∗2∂(τ1A1)
< 0, we expect to observe that:
Hypothesis 1 : As industry generates more local tax revenue, local politicians allocate more
land quotas for industrial use and consequently fewer land quotas for commercial and residential use.
(2)∂k∗1
∂(ατ2A2)< 0,
∂k∗2∂(ατ2A2)
> 0, we expect to observe that:
Hypothesis 2 : As the service sector generates more local tax revenue, local politicians allocate
fewer land quotas for industrial use and consequently more quotas for commercial and residential
use.
(3) Given that the land conveyance fee generated per unit of land used for commercial and
residential purposes is higher than that from land for industrial use, i.e., (s1 − s2) < 0, we get
∂k∗1∂d < 0,
∂k∗2∂d > 0, we expect to observe that:
Hypothesis 3 : Local politicians with longer time horizons allocate more land quotas for industrial
16
use and consequently fewer quotas for commercial and residential use.
4 Data and Measurement
4.1 Sample
I compiled an original dataset for a probability sample of 120 prefectural-level cities to test the
hypotheses. These 120 cities are a probability sample constructed by the World Bank in its survey
of “China Governance, Investment Climate, and Harmonious Society: Competitive Enhancement
for 120 Cities in China” in 2005 (World Bank Report No. 37759-CN, 2006). Cities in China vary
in administrative levels: there are four provincial-level municipalities, 15 deputy provincial cities,27
268 prefectural-level cities, and 368 county-level cities (China City Statistical Yearbook, 2009). The
survey selects cities at the prefectural level and above. In particular, all 19 cities from the first two
categories are included in the sample. The 120 cities are from all provinces except Tibet. For each
province, the capital city is included. The inclusion of additional cities for a particular province
depends on provincial GDP. A map in Figure 8 shows where the 120 cities are located: 287 cities at
the prefectural level and above are represented by dots, and the 120 cities included in the sample
are highlighted. The 120 cities account for 70-80 percent of China’s GDP in 2005, when the survey
was conducted.
[Figure 8 about here]
4.2 Data Compilation
Data on the 120 cities comes from official Chinese publications and sources. Tax revenue data comes
from Financial Statistical Material for Prefectures, Cities, and Counties Nationwide, issued by the
Ministry of Public Finance. China Land and Resources Statistical Yearbook, issued by the Ministry
of Land and Resources, has land revenue data disaggregated at the prefectural level. Economic
data (e.g., GDP per capita, city population, and economic structure) comes from City Statistical
Yearbook. Data on political elites is compiled from career histories of city leaders, including both
27The four provincial-level municipalities are Beijing, Tianjin, Shanghai, and Chongqing. The 15 deputy provinciallevel cities are Haerbin, Changchun, Shenyang, Dalian, Jinan, Qingdao, Nanjing, Hangzhou, Ningbo, Xiamen,Guangzhou, Shenzhen, Wuhan, Chengdu, and Xi’an.
17
Party secretaries and mayors. Data from the various sources cited above is collected for the
probability sample of 120 cities to form a new original dataset.
4.3 Measurement of Key Variables
Dependent Variable: Land quota allocation between industry and services is the dependent
variable. As discussed in the previous section, industry occupies land for industrial purposes,
whereas tertiary industry occupies land for commercial and residential purposes. Ideally, we want
to know how many land quotas are distributed to each sector. With this option not available at the
city level, land allocation between the two sectors is proxied by land allocation between nonmarket
and market transactions, because land revenue differentiation by sectors is caused by the way that
land leases are transacted. Local governments have four methods to lease urban construction land
use rights: one non-market approach of arranging one-on-one meetings with potential land users
to negotiate the land lease, and three market approaches of bid invitation (zhaobiao), auction
(paimai), and quotation (guapai). In any of the last three approaches, the price of land leases is
driven by market forces, thus is substantially higher than the negotiated price. Most industrial
land use rights are leased out by nontransparent negotiation between local officials and firms. In
2006, 97 percent of land used for industry was leased by negotiation, and nearly 80 percent of land
used for commercial and residential purposes was leased by market approaches (i.e., bid invitation,
auction, and quotation combined).28
[Figure 10 about here]
I use the ratio of land transacted by negotiation to the total amount of land available to
measure land quotas allocated to industry. Since a ratio is bounded between 0 and 1, a logit
transformation is employed to convert the bounded dependent variable into an unbounded one.
Figure 10 plots both histogram and density for the logit transformation of land quota allocation
ratio. The logit transformation of the dependent variable is normally distributed centering at the
28This proxy is not a good measurement after 2007, when the central government required state-owned constructionland to be leased out using market approaches only. See State Council, Circular of the State Council on IntensifyingLand Control, 31 August 2006, Article 5; Ministry of Land and Resources, Provisions on the Assignment ofState-owned Construction Land Use Right through Bid Invitation, Auction, and Quotation, 28 September 2007,Article 4. In 2007, 74 percent of land used for industry was leased out by negotiation. In 2008, this ratio dropped to17 percent.
18
mean of 0.7, meaning that an average of 65 percent of land quotas are allocated to industry.
Independent Variables: Three sets of variables are of theoretical interest: economic
determinants, political determinants, and the constraint of land quotas.
Measurement of τ1A1: The tax rate in China is determined hierarchically. Under the TSS,
the central government sets the tax sharing ratios with the province only, leaving the latter with
substantial flexibility in determining the tax sharing ratios with their prefectures, and the prefecture
with its counties, and so on. As a result, the tax rate τ1 is expected to vary across cities. Ideally
for A1, we want to have some measure on firm productivity aggregated at the city level. Data for
neither variable is publicly accessible for all 30 provinces covered in the survey. Hence, I identify
VAT to measure τ1A1 indirectly. VAT is shared between the central and local governments at a ratio
of 3:1.29 It is a major local tax revenue source paid primarily by industry on the added value derived
from production. If we assume τ1 to be the value-added tax rate, an increase in both τ1 and A1 is
expected to produce higher VAT retained by local governments. Arguably, land, as a productive
input, may affect VAT in some indirect way, so that the measure produces an endogeneity problem.
Empirically however, land value-added tax is an independent category and paid separately from the
VAT under the existing TSS system. To reduce the potential endogeneity, I construct a variable
called VAT ratio measured by the proportion of VAT in the total local budgetary revenue.
After the central government collects local revenue (e.g., 75 percent of VAT), it redistributes
the collected revenue to local governments in the form of intergovernmental fiscal transfers. The
intergovernmental transfer in China is primarily comprised of tax rebates and earmarked transfers.
The proportion of tax rebates in total transfers has been declining whereas the proportion of
earmarked transfers has been increasing over time. For instance, in 1996, tax rebates accounted
for 72 percent of intergovernmental transfers, and this ratio declined to 45 percent in 2001 (World
Bank 2002, 19). Tax rebates are furthered categorized into tax rebates for income taxes, and tax
rebates for VAT and consumption taxes. To include the VAT in the form of intergovernmental
transfers, I construct a variable called tax rebates ratio, measured by the proportion of tax rebates
for VAT and consumption tax in total intergovernmental fiscal transfers received by each city. The
29This VAT shared ratio means that the central government takes 75 percent of VAT, whereas the 25 percent isshared by all subnational government units, i.e., province, prefecture, county, and township. The shared ratio isdetermined hierarchically.
19
tax rebates for income taxes are not included in the numerator because they are contributed by
both industry and services.
Measurement of ατ2A2: The major tax generated by tertiary industry is business tax.
Consistent with the VAT ratio measure, I construct a variable called business ratio measured
by the proportion of business tax in total local budgetary revenue.
Measurement of d: Recall that d represents the discount factor, having an inverse relationship
with time horizon: a local politician with a long time horizon values the future more than one with
a short time horizon, thus having a smaller discount factor. Clague et al. (1996) use the age of an
autocratic regime to measure autocratic time horizon. Similarly, I use the tenure in office of a local
politician to measure his time horizon. For each city, I consider the two most important leading
positions in the analysis: party secretary and mayor. I construct two variables, CCP secretary
tenure and mayor tenure, to measure how many years a local party secretary or a mayor has been
holding his current position, as of the year 2005 when the World Bank survey was conducted. Since
both positions are limited to two terms with each term 5 years, both variables are discrete, ranging
from 1 to 10. As a local politician increases his tenure in office, his time horizon shortens. Such a
measure, however, fails to take the term constraint into account: an official who starts his second
term may have a longer time horizon than one who is close to the end of his first term but will
not serve a second term. Another concern arises when politicians have not completed their terms
but know they will be taking different positions elsewhere, causing their time horizon to reach the
smallest value long before their full term ends. Thus, the tenure effect on the time horizon may
be nonlinear. To address this concern, I add a quadratic form of tenure in office for both party
secretaries and mayors.
Normally, a party secretary and a mayor in a city are two different individuals. Two out of
120 cities surveyed (Quanzhou in Fujian and Xianyang in Shaanxi) had the same person serving
concurrently as party secretary and mayor. Since the appointment time for the same person with
respect to the two positions differs, thereby generating different values for the time horizon, I treat
the concurrent party secretary and mayor as two different persons.
Measurement of k: Recall that k is the constraint in the model. In the empirical analysis,
I construct a variable called land quota per capita measured by the ratio of total amount of land
available for lease to the total population for each city. Annual land quotas are distributed from
20
above at the beginning of each year. Because the time horizon is measured by year, it is possible
that political leaders who start their appointment in year 2005 are unable to be involved in the
decision-making on land quota distribution for that year. For this consideration, the land data is
measured in year 2006.
Control Variables: I included a number of controls for factors that might be expected to
affect local land quota allocation. To capture the existing economic structure, the ratio of the
secondary sector in GDP is included. To capture the fiscal gap between formal budgetary revenue
and expenditure facing local governments, I construct the variable budget deficit, which follows the
measure by Lorentzen, Landry, and Yasuda (2011), using the ratio of local expenditure to local
budgetary revenue. A local government runs a surplus when the ratio is smaller than 1, and a
deficit when the ratio is greater than 1. The larger the deficit, the larger the ratio. To capture
city characteristics, I included the following variables: GDP per capita, urbanization measured by
the ratio of urban population to total population, and a city dummy for the 19 cities at provincial
and deputy-provincial levels. Another variable that may have an impact on the decision making
process of local politicians is their age. Local politicians will be removed from office when they get
close to the mandatory retirement age. To take the retirement regulation into account, I construct
variables distance to retirement for both party secretaries and mayors, measured by, as of 2005,
how many years until they reach age 65 for local politicians at provincial-level municipalities and
age 60 for those at deputy-provincial and prefectural-level cities.30
To match the World Bank survey conducted in 2005, all variables are measured for year 2005
with only two exceptions related to land quotas. These two exceptions are land quota allocation
(i.e., the dependent variable) and the land quota constraint (i.e., k), which are measured in year
2006 for the reasons explained above. Table 3 presents summary statistics of the key variables.
The definition and data source for each variable are detailed in Appendix B.
[Table 3 about here]
30Retirement age varies with bureaucratic rank. See State Council, Interim Regulations on the State Civil Servant,Aug 14, 1993, Article 78.
21
5 Analysis
5.1 Findings
Table 4 reports the OLS regression results for land quotas allocated to industry. To allow for
heteroskedasticity across observations, regressions are estimated with robust standard errors. I
begin my analysis with the estimates that regress land quota allocation against all independent
and control variables identified above, except the quadratic form of tenure in office, and report the
results in column (1). The coefficient of CCP secretary tenure is negative and significant at the 5%
level, meaning that fewer land quotas are allocated for industry as party secretaries increase their
time in office. The coefficient of mayor tenure, however, is not statistically significant. Column (2)
reports the estimates when the quadratic form of tenure in office is included in the regression for
both party secretaries and mayors. In comparison with the regression estimates reported in column
(1), this regression produces better results: not only do the coefficients of the variables significant
in regression (1) remain significant, but the significance level of two variables (tax rebates ratio and
CCP party secretary tenure) also improves. More importantly, the coefficient of the quadratic form
of party secretary tenure in office is statistically significant at the 1% level, indicating a nonlinear
relationship between tenure in office and land quota allocation. Thus, regression (2) is preferred
and is treated as the baseline model.
[Table 4 about here]
I begin my interpretation of regression (2) estimates with the economic determinants, followed
by political determinants, the land quota constraint (i.e., k in the model), and control variables.
The coefficients of all three economic determinants (i.e., VAT ratio, tax rebates ratio, and business
tax ratio) are statistically significant at the 5% level. The positive coefficients of VAT ratio and tax
rebates ratio indicate that more land quotas are allocated to industry as local tax revenue relies
more on VAT, consistent with the first hypothesis. A calculation of marginal effect shows that a 1
percent increase in the ratio of VAT to the total budgetary revenue is associated with an average
increase of 0.7 percent in the ratio of land quotas for industry to the total amount of land quotas
(for simplicity, I will refer to this as the industrial land quota ratio), and a 1 percent increase
in the ratio of tax rebates to the total intergovernmental transfer is associated with an average
22
increase of 0.44 percent in the industrial land quota ratio.31 The negative coefficient of business
tax ratio indicates that fewer land quotas are distributed to industry, as the local tax revenue relies
on business tax to a greater extent, consistent with the prediction of the second hypothesis. A 1
percent increase in the ratio of business tax to local budgetary revenue is associated with a decrease
in the industrial land quota ratio by an average of 0.63 percent.
The regression confirms a negative relationship between the time horizon of local politicians and
land quota allocation: a politician with a short time horizon is interested in the more intermediate
revenue associated with tertiary industry, and thus he is likely to invest fewer land quotas in
industry. The time horizon of local politicians varies with their time in office. The coefficients of
both tenure and its square term for party secretaries are both statistically significant at the 1%
level, together with their signs, indicating the relationship between tenure in office and the industrial
land quota ratio is parabolic with its vertex oriented download at the tenure of 5.4 years.32 This
estimation is consistent with theoretical prediction. Recall the institutionally stipulated tenure
length for one term is 5 years, and our estimated tenure that produces shortest time horizon is 5.4
years. That is, as a local politician increases his time in office, fewer land quotas are allocated to
industry. The industrial land quota ratio reaches its minimum when the politician stays in office
for 5.4 years. After that, if he continues in his position, more land quotas will be distributed to
industry.
As in regression (1), the tenure of mayors does not have an impact on local land quota allocation.
Neither the coefficient of mayor tenure nor its quadratic form is statistically significant. The party
secretary and mayor are the most important two local political elites, but the former, not the latter,
has a significant impact on local land quota allocation. There are two plausible explanations. First,
the decision-making power on land quota allocation is concentrated in the party secretary, not
mayors. In an interview with a party secretary of a district in Wenzhou, the party secretary said,
“Our government is now operating like a large corporation. The party secretary is the chairman
31Note that the regressand is a logit transformation because the dependent variable, denoted as y, is a ratio boundedbetween 0 and 1. Denote the logit transformation as z = ln( y
1−y ), independent variable as x, and coefficient as β.
To get the marginal effect of y on x, I calculate as follows: ∂y∂x
= ∂z∂x/ ∂z∂y
= βy(1 − y). This expression is nonlinear, I
then use the mean, y, to calculate the average marginal effect, i.e., ∂y∂x
= βy(1 − y), where y = 0.645, shown in Table3, and all βs are estimated by the regression.
32Denote the variable of secretary tenure as x1 and its coefficient is a, and the coefficient for its quadratic form isb. Thus, z = ax1 + bx21. To find the minimum,I derive the first order condition, set it equal to 0, and solve for x1,i.e., ∂z
∂x= a+ 2bx1 = 0, thus, x1 = − a
2b= − −0.475
2×0.044= 5.4
23
of the board and the mayor is the CEO” (Interview ZJ12160109). The quote suggests there might
exist a division of labor between the two positions, despite both being crucial local figures. In an
examination of government transparency using city level data in China, Lorentzen, Landry, and
Yasuda (2011) found that only the mayor but not the party secretary matters for improving local
government transparency on environmental regulations. This suggests that the division of labor
between party secretary and mayor depends on issue area.
An alternative explanation is that the measurement of tenure in office for mayors fails to capture
their time horizons. Despite the formal regulation on tenure length (i.e., five years), the CCP
has nonetheless acted informally to reduce the tenure of chief government executives (not party
secretaries), resulting in their actual time served in office being shorter than the full term. Using a
panel data on city mayors from 1990-2001, Landry (2008) found that the average length of mayor
tenure has been steadily shrinking from an 3.2 years in 1990 to a mere 2.5 years by 2001. In an
examination of county chief executives from 1998 to 2002, Guo (2009) shows that 23 percent of
county chief executives in their fourth year are promoted to be party secretaries locally or elsewhere.
In the situation where the gap between actual tenure in office and institutionally stipulated tenure
is large, some mayors may update their belief on actual time in office and alter their behavior
accordingly, while others may be uncertain about whether they will be lucky enough to be promoted
long before they complete their terms, and therefore their time horizons and behavior will remain
as predicted in my initial discussion. If both scenarios occurred, we would expect to observe two
peaks at which the time horizon of mayors reaches local minimum: one at the time when mayors
are expected to be taking another position in the middle of their first term and one at the end of
the institutionally designated tenure. A simple linear regressor and its quadratic form in regression
(2) fail to capture such a relationship.
The coefficient of land quota constraint is statistically significant at the 1% level. The positive
sign indicates that more land quotas are allocated to industry as land quota availability increases.
As land quotas increase by 1 square kilometer per 1000 persons, the industrial land quota ratio
increases by an average of 0.23 percent. Among all the controls, only the ratio of industry to local
GDP has a small impact on land quota allocation. A 1 percent increase in the ratio of industry
to local GDP is associated with a decrease in the industrial land quota ratio by an average of 0.01
percent. Surprisingly, budget deficits do not matter to local land quota allocation. A plausible
24
explanation might be that the formal budget deficit cannot capture the real fiscal pressure facing
local politicians because of the alternative revenue sources from intergovernmental transfers and
extrabudgetary revenue.
5.2 Robustness Check
I conduct three sensitivity tests to check that the major findings are robust to alternative
specifications. Since the mayor does not have an impact on land quota allocation in regressions
(1) and (2), all variables related to mayors are dropped in the sensitivity tests. The variable of
party secretary tenure in my data has a mean of 3.3 years and standard deviation of 1.64, as
shown in Table 3, meaning that not many party secretaries were serving their second term at the
time when the survey was conducted. It is possible that party secretaries in their second terms
behave differently from those serving their first term. To address this concern, a dummy variable
second term indicator is constructed for party secretaries who remain in their positions more than
five years. Column (3) reports the regression estimates. The coefficients for both economic and
political determinants remain statistically significant with the same signs. The values of these
coefficients are similar to those in regression (2). With slight changes in the value of coefficients,
the age at which the time horizon reaches the minimum estimated in regression (3) is 4.21 years,
about one year smaller than the estimation in regression (2).
A simple cross-sectional analysis like the one conducted here runs a risk of omitted variable bias.
To address this, I include additional variables that might be expected to affect land quota allocation
by party secretaries. A city that receives more tourists is expected to have a greater demand for
tertiary industry (e.g., hotels, restaurants), thereby more land quotas allocated to tertiary industry.
I include a variable called tourist ratio measured by the proportion of tourists received to the total
local population as of 2005. An additional two variables are added to capture the possibility that
individual characteristics of party secretaries may shape their preferences on one sector over the
other in land quota allocation: education level and a dummy variable indicating if a party secretary
was locally promoted to his current position. A local politician with better education may have a
longer time horizon than the one who is less educated. A local politician who was locally promoted
may care more about the local economy in the long run than one promoted from elsewhere. The
results reported in column (4) indicate that none of these additional variables matters for land
25
quota allocation. The signs, statistical significance, and values of the coefficients of economic and
political determinants are all consistent with the estimates in regression (2).
In previous analyses, the effect of age is measured by distance to mandatory retirement age.
This measure assumes a linear relationship between age and land quota allocation. However, the
age effect may be nonlinear, similar to the tenure effect. To allow a nonlinear relationship, I replace
the linear regressor of distance to retirement by two variables: age as of 2005 and its quadratic
form. Regression estimates are reported in column (5). The coefficient of age is statistically
significant at the 1% level, but it is not robust. It is no longer statistically significant once I drop
four provincial-level municipalities from the sample because the retirement age for officials at this
level is five years greater than that for the rest of the population, as shown in column (6). All
the coefficients of economic and political determinants keep their signs and statistical significance.
Moreover, the coefficient of second term indicator is now statistically significant at the 5% level.
The negative sign indicates that a local party secretary in his second term is likely to allocate fewer
land quotas to industry than one in his first term. Overall, the main findings that both revenue
structure and the time horizon of local politicians matter for local land quota allocation are robust
to several sensitivity tests.
6 Conclusion
The 1994 fiscal reform had an important impact on local revenue structure. It reduced local
budgetary revenue without adjusting expenditure responsibilities, leaving local governments in
poor financial condition. The fiscal pressure facing local governments and the distorted land market
made leasing urban construction land use rights an increasingly popular source of revenue to help
local politicians fulfill unfunded mandates. The incentive for local politicians to expropriate land
however, is constrained by the land quota system, a system that the central government imposed,
attempting to preserve land for food security and social stability. This quota system limits the total
amount of land that local politicians can expropriate for urban construction, but still leaves local
politicians enough autonomy to determine how to distribute quotas across economic sectors. Taking
advantages of sectoral differences in local revenue contribution, this paper investigates how local
politicians balance their land quotas, a scarce resource, between industry and services. It argues
26
that both the local revenue structure and the time horizon of local politicians have an impact on
local land quota allocation. With systematic analysis of 120 cities randomly sampled in 2005, the
paper finds that more land quotas are distributed to industry when the local revenue base relies
more on VAT and less on business tax, when local CCP secretaries (but not mayors) have long
time horizons, and when the locality is assigned more quotas. These findings are robust to several
sensitivity tests.
Some scholars argue that the 1994 fiscal reform motivated local governments to shift their
economic development strategy from promoting industry to services, because local politicians do
not share with the central government their business tax generated by services, but must surrender
75 percent of VAT, generated by industry, to the central government (F. Zhou 2006; X. Jiang et
al. 2007). This argument is consistent with the development of local tax structure over time: the
proportion of business tax in local budgetary revenue has been increasing from 28 percent in 1994
to 32 percent in 2008, whereas the proportion of VAT has been declining from 25 percent to 19
percent within the same period. However, the data aggregated at the national level disguise the
subnational variation in the local tax base. Using sectoral allocation of land quotas as a case, this
paper provides both an economic and a political logic, explaining the subnational variation in local
revenue extraction strategy and refining the argument on the impact of the fiscal reform on local
economic development.
My data show that land quota allocation is always a mix of industry and services. Why would
not we observe the situation where one industry owns all quotas and the other gets none? To manage
a complicated business like land quota allocation, local politicians have many considerations other
than their time horizon and the local revenue structure, highlighted in this paper. An official who
was in charge of a national economic development zone in Zhejiang explained, “We cannot earn
money by selling land [use rights] to manufacturing firms, and oftentimes we lose money because the
profit from land sales is not enough for us to cover the cost of building infrastructure to attract them
to invest here. What we expect to gain from them is not land profits, but their tax contribution in
the long run. . . .Our loss from land sales to manufacturing firms can be offset by the profits gained
from real estate” (ZJ03190110). Local politicians must balance their budgets (Budget Law, Article
3). Therefore, even local politicians with the longest time horizon, who prefer the development of
manufacturing industry, must allocate land quotas to tertiary industry. Similarly, those with the
27
shortest time horizon, who have a strong preference for windfall land profits from real estate, are
forced to allocate quotas to the manufacturing industry so as to fulfill mandates like budgetary
revenue and FDI attraction.
Finally, although this paper focuses on land, the economic and political determinants derived
from quota allocation may have broad applications. In China, the state still maintains monopoly
control over many resources (e.g., mining), not only land. It is reasonable to expect that politicians
will take the revenue structure and their time horizon into account in allocating the resources they
control, as long as they care about revenue.
28
Figure 1: Value added by industry as a share of GDP of transitional countries, 1990-2009
3035
4045
50in
dust
ry s
hare
of G
DP
(%)
1990 1995 2000 2005 2010Time
EE_Baltics CISChina Russia
Source: World Bank, World Development Indicators, available at http : //databank.worldbank.org.
Figure 2: Value added by services as a share of GDP of transitional countries, 1990-2009
3040
5060
70se
rvic
e sh
are
of G
DP
(%)
1990 1995 2000 2005 2010Time
EE_Baltics CISChina Russia
Source: World Bank, World Development Indicators, available at http : //databank.worldbank.org.
29
Figure 3: Industry (value added) share of GDP, worldwide, 2009
China
AlgeriaAngola Azerbaijan
Congo, Rep.
Equatorial Guinea
GabonGuinea
India
Russia
Saudi Arabia
Trinidad and TobagoTurkmenistan
020
4060
8010
0In
dust
ry, v
alue
add
ed (%
of G
DP)
0 10000 20000 30000 40000
GDP per capita, PPP
Source: World Bank, World Development Indicators, available at http : //databank.worldbank.org.
Figure 4: Service (value added) share of GDP, worldwide, 2009
China
AlgeriaAngola Azerbaijan
Congo, Rep.
Equatorial Guinea
Gabon
Guinea
India
Russia
Saudi Arabia
Trinidad and Tobago
Turkmenistan
020
4060
80Se
rvic
e, v
alue
add
ed (%
of G
DP)
0 10000 20000 30000 40000
GDP per capita, PPP
Source: World Bank, World Development Indicators, available at http : //databank.worldbank.org.
30
Figure 5: GDP Growth Rate and Arable Land Occupied by Construction in China, 1985-2008
0
1000
2000
3000
4000
5000
6000
0 2
4 6
8 10 12 14 16
1985
19
86
1987
19
88
1989
19
90
1991
19
92
1993
19
94
1995
19
96
1997
19
98
1999
20
00
2001
20
02
2003
20
04
2005
20
06
2007
20
08
Ara
ble
land
Ocu
pied
by
Con
stru
ctio
n (1
000
mu)
GD
P G
row
th R
ate
(%)
gdp growth rate Arable land occupied by construction
Source: On GDP growth rates, World Bank, World Development Indicators, available at http ://databank.worldbank.org. On arable land data, China Land and Resources Statistical Yearbook, 2006, p. 19;2009, p.19; Hu et al., Analysis of Land Use in Guanzhou, 2009, p. 48; China Rural Statistical Yearbook, 1991, p.235; 1993, p. 229; 1995, p. 70; 1996, p. 54.
Figure 6: Subnational Budgetary Revenue and Expenditure, 1978-2008
0% 10% 20% 30% 40% 50% 60% 70% 80% 90%
1979
1980
1981
1982
1983
1984
1985
1985
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
PERCENTAGE
YEAR
Ratio of local budgetary revenue to the total budgetary revenue
Ratio of local budgetary expenditure to the total budgetary expenditure
Source: China Fiscal Yearbook, 2009, pp. 486-89.
31
Table 1: Cadre Evaluation of a Prefecture-level City in Guangdong Province, 2009
Evaluation Indicator Baseline Score Bonus/Subtraction Score Score Calculation
GDP growth rate (%) 13 Rank 1 to 5 and obtain additionalpoints consisting with the ranking:1.3, .91, .65, .52, .39
1. Obtain the baseline score if thetarget is reached.2. Obtain the baseline scoresubtracting the additional score ifthe target is not reached butfulfilled at least 90%.3. Score 0 if the target is fulfilledless than 90%.4. Obtain baseline score plusbonus score if the target isover-fulfilled.
Local budgetary revenuegrowth rate (%)
15 Rank 1 to 5 and obtain additionalpoints consisting with the ranking:1.5, 1.05, .75, .60, .45
Local fixed-assetinvestment growth rate(%)
8 Rank 1 to 5 and obtain additionalpoints consisting with the ranking:.80, .56, .40, .32, .24.
Real FDI growth rate (%) 7 Rank 1 to 5 and obtain additionalpoints consisting with the ranking:.70, .49, .35, .28, .21
GDP energy consumptionreduction rate (%)
13 No rankingGet full score if the target isfulfilled and 0 otherwise.
SO2 emission 10 No rankingChemical oxygen demand(COD) emission
7 No ranking
Development of Hi-techZone
7 Rank 1 to 5 and obtain additionalpoints consisting with the ranking:.70, .49, .35, .28, .21
Each indicator is further dividedinto many categories measured bysub-indicators. The indicator isscored 0 as long as one of itssub-indicators is not fulfilled.
Social security 10 No rankingSocial safety control (I):Production safety control
5 Rank 1 to 5 and obtain additionalpoints consisting with theranking:.50, .35, .25, .20, .15
Social safety control (II):Fire accident control
5 No ranking
Total Score 100
Source: Internal government document obtained from fieldwork, 2010.Note:This prefecture-level city is comprised of five districts. Thus, the ranking is from one to five.The first four indicators are ranked based on the formula: actual performance value−target value
target value× 100%. Indicators
#2, #3, and #4 are also ranked in absolute term, i.e., (actual performance value − target value), the numeratorof the previous formula. That is, GDP final score is a sum of baseline score and one additional score, while the finalscore for indicators 2-4 constitute a baseline score and two additional scores. The indicators of development of hi-techdevelopment zone and production safety control are further divided into several sub-indicators. The two indicatorsare ranked according to the sum of their sub-indicator scores.
32
Figure 7: Budget Deficit against Local GDP of 120 Cities, 2005
12
34
56
expe
nditu
re/re
venu
e ra
tio
0 20000 40000 60000 80000local GDP per capita in 2005
Source: Author’s dataset. Data on local GDP per capita is compiled from China City Statistical Yearbook, 2006.Data on local budgetary revenue and expenditure is compiled from Financial Statistical Material for Prefectures,Cities, and Counties Nationwide, 2005.
Figure 8: A Sample of 120 Cities
33
Table 2: Local Tax Contribution, by Sectors, 2003-2008
Sectors VAT Business Tax Income TaxMining × ×Manufacturing × ×
Industry Production and supply of electricity, gas, and water × ×Construction × ×Transport, Storage, and Post × ×Wholesale and retail* × ×Finance × ×
Service Information transmission, computer, and software × ×Business services × ×Real estate × ×Others × ×
Source: China Tax Yearbook.Note: *Data from the China Tax Yearbooks in years 2001 and 2002 indicates that the wholesale and retailindustry pays both VAT and business tax, and the yearbooks after 2003 shows that it pays VAT only.
Figure 9: Sectoral Difference in land revenue (10,000RMB/hectare), 2004-2008
0
50
100
150
200
250
300
350
400
450
2004 2005 2006 2007 2008
industrial
commercial and residential
Source: China Land and Resources Statistical Yearbooks, 2005-2009.
34
Figure 10: Histogram and Density for Land Allocation Ratio, 2006
0.1
.2.3
.4.5
Hist
ogra
m a
nd d
ensit
y
-2 0 2 4
logit transformation of industrial land allocation ratio
Table 3: Summary Statistics of Key Variables
Variable Obs Mean Std. Dev. Min MaxLand quota allocation 118 .645 .181 .113 .962Land quota allocation (logit) 118 .700 .912 -2.058 3.228VAT ratio 120 .178 .052 .077 .312Tax rebates ratio 120 .239 .124 .0193 .559Business tax ratio 120 .239 .073 .087 .470CCP secretary tenure 120 3.325 1.641 1 9Mayor tenure 120 3.342 1.585 1 10Land quota per capita 119 2.593 2.491 .098 15.272Ratio of secondary industry in GDP 120 48.499 9.092 27 85.92Secretary distance to retirement 120 7.833 3.8272 -1 17Mayor distance to retirement 120 9.35 4.272 1 19Budget deficit 120 1.834 .765 .811 5.749GDP per capita (log) 120 9.750 .624 8.340 11.146Urbanization 120 .404 .199 .116 1City level 120 .158 .367 0 1
35
Table 4: OLS Estimates for Land Quota Allocated for Industry
Dependent variable=land quota allocated to industry in logit transformation
(1) (2) (3) (4) (5) (6)
VAT ratio 3.014** 3.041** 3.338** 3.216** 3.554*** 3.842***(1.519) (1.493) (1.326) (1.327) (1.354) (1.334)
Tax rebates ratio 1.721* 1.913** 2.229*** 2.399*** 1.991** 1.811**(0.906) (0.880) (0.784) (0.796) (0.793) (0.849)
Business tax ratio -2.741** -2.750** -2.116* -2.066* -2.235* -2.333*(1.211) (1.188) (1.081) (1.094) (1.178) (1.235)
CCP secretary tenure -0.132** -0.475*** -0.614*** -0.581*** -0.651*** -0.863***(0.056) (0.139) (0.192) (0.187) (0.176) (0.219)
CCP secretary tenure square 0.044*** 0.073** 0.068** 0.082*** 0.117***(0.016) (0.030) (0.028) (0.026) (0.034)
Secretary second term indicator -0.848 -0.794 -0.957* -1.239**(0.570) (0.565) (0.550) (0.573)
Mayor tenure -0.019 0.153(0.071) (0.158)
Mayor tenure square -0.019(0.016)
Land quota per capita 0.112*** 0.100*** 0.084** 0.098** 0.078* 0.079*(0.042) (0.038) (0.041) (0.044) (0.041) (0.042)
Industry to GDP ratio -0.030** -0.033** -0.027** -0.027** -0.028** -0.025*(0.014) (0.013) (0.013) (0.013) (0.013) (0.013)
Secretary distance to retirement -0.027 -0.028 -0.029 -0.023(0.027) (0.027) (0.025) (0.021)
Mayor distance to retirement 0.013 0.017(0.019) (0.019)
Secretary age as of 2005 0.586* 0.461(0.341) (0.377)
Secretary age square -0.005 -0.004(0.003) (0.004)
GDP per capita(log) -0.211 -0.088 -0.167 -0.156 -0.117 -0.116(0.400) (0.407) (0.367) ( 0.403) (0.362) (0.372)
Urbanization 0.883 0.850 0.792 0.800 0.833 0.814(0.809) (0.805) (0.785) (0.815) (0.772) (0.789)
Budget deficit -0.137 -0.067 -0.048 -0.042 -0.039 -0.021(0.221) (0.206) (0.170) (0.174) (0.164) (0.162)
City dummy -0.161 -0.274 -0.235 -0.297 -0.182 -0.114(0.228) (0.225) (0.223) ( 0.226) (0.230) (0.240)
Tourist ratio -0.020(0.061)
Secretary education -0.024(0.100)
Secretary local promotion 0.077(0.143)
Constant 4.154 3.182 3.916 3.800 -12.452 -9.112(3.874) (3.937) (3.298) (3.531) (9.398) (10.279)
Observations 118 118 118 117 118 114R2 0.343 0.382 0.387 0.405 0.402 0.415
Robust standard errors are in parenthesis.* p < 0.1, ** p < 0.05, *** p < 0.01
36
Appendices
A Interviewee List
The interviewee code is organized as follows. The two letters indicate the provinces in which the
interview was conducted. The first four digits after the letters indicate the month and date of the
interview. The next digit indicates if the interview subject was a group or an individual (coded
1 for group and 0 for individuals). This is followed by a digit indicating the order of group or
individual interviewed on that day. The last two digits indicate the year in which the interview was
conducted. To make interview subjects unidentifiable, I release only the name of the provinces,
rather than more specific locality information. The table below identifies the interview subjects by
level (where relevant) and type of workplace institution.
Code Locality Level Interviewee Institutions
ZJ12160109 Zhejiang Urban district Party secretary Party committeeGD01311110 Guangdong Prefecture Group of government
officialsBureaus of Land and Resources, PublicFinance, Development and Reform,Urban Planning, Hi-tech EconomicZone management committee
GD03040110 Guangdong Township Section chief Economic development zoneGD03040210 Guangdong Township Entrepreneur Government sponsored land
development companyGD03050110 Guangdong Village Party secretary Village party committeeZJ03190110 Zhejiang Deputy
provincial citySection chief National Economic development zone
ZJ03190210 Zhejiang Deputyprovincial city
Director National economic development zone
ZJ03290110 Zhejiang N/A Researcher UniversityJS04020110 Jiangsu County-level city Director National economic development zoneJS04060110 Jiangsu Village Party secretary Village party committeeJS04071110 Jiangsu County-level city Group of government
officialsHi-tech Development zone
ZJ04120110 Zhejiang Prefecture Director Bureau of AgricultureZJ04130110 Zhejiang Prefecture Director Provincial economic development zoneZJ04160110 Zhejiang County-level city Director County party committeeJZ04160210 Zhejiang Prefecture Department chief Bureau of Land and ResourcesJZ04190110 Zhejiang County-level city Director Provincial economic development zoneJZ04190210 Zhejiang Township Director Economic and Trade OfficeJZ04200110 Zhejiang County-level city Vice mayor GovernmentJZ04200210 Zhejiang County-level city Deputy director Provincial economic development zoneCQ04250110 Chongqing N/A Researcher Research InstituteCQ04280110 Chongqing Province Director Bureau of Land and ResourcesCQ05060110 Chongqing Province Vice secretary general Party committeeCQ05060210 Chongqing Province Section chief Bureau of Reform and DevelopmentCQ05110110 Chongqing Urban district Director Economic development zoneCQ05121210 Chongqing Urban district Group of government
officialsBureau of Land and Resources
37
BD
efi
nit
ion
an
dS
ou
rces
of
Vari
ab
les
Vari
able
Defi
nit
ion
Sourc
e
Indust
rial
land
quota
Logit
transf
orm
ati
on
of
the
rati
oof
land
transa
cted
by
neg
oti
ati
on
toto
tal
am
ount
of
land
available
Chin
aL
and
and
Res
ourc
esSta
tist
ical
Yea
rbook
VA
Tra
tio
Rati
oof
VA
Tto
the
tota
lbudget
ary
reven
ue
Fin
anci
al
Sta
tist
ical
Mate
rial
for
Pre
fect
ure
s,C
itie
s,and
Counti
esN
ati
onw
ide
Busi
nes
sta
xra
tio
Rati
oof
busi
nes
sta
xto
the
tota
lbudget
ary
reven
ue
Fin
anci
al
Sta
tist
ical
Mate
rial
for
Pre
fect
ure
s,C
itie
s,and
Counti
esN
ati
onw
ide
Tax
rebate
sra
tio
Rati
oof
tax
rebate
sto
the
tota
lin
terg
over
nm
enta
ltr
ansf
erF
inanci
al
Sta
tist
ical
Mate
rial
for
Pre
fect
ure
s,C
itie
s,and
Counti
esN
ati
onw
ide
CC
Pse
cret
ary
tenure
Ten
ure
of
CC
Pse
cret
ary
inyea
rsC
om
piled
by
the
auth
or
from
vari
ous
sourc
esM
ayor
tenure
Ten
ure
of
may
or
inyea
rsC
om
piled
by
the
auth
or
from
vari
ous
sourc
esL
and
quota
per
capit
aT
he
rati
oof
tota
lam
ount
of
land
toto
tal
popula
tion
Chin
aL
and
and
Res
ourc
esSta
tist
ical
Yea
rbook
Rati
oof
Indust
ryin
GD
PR
ati
oof
seco
ndary
indust
ryin
GD
PC
hin
aC
ity
Sta
tist
ical
Yea
rbook
Sec
reta
rydis
tance
tore
tire
men
tD
ista
nce
toth
ere
tire
men
tage
for
CC
Pse
cret
ari
es(i
.e.,
the
requir
edre
tire
men
tage−
the
age
as
of
2005)
Com
piled
by
the
auth
or
from
vari
ous
sourc
es
May
or
dis
tance
tore
tire
men
tD
ista
nce
toth
ere
tire
men
tage
for
CC
Pse
cret
ari
esC
om
piled
by
the
auth
or
from
vari
ous
sourc
esB
udget
defi
cit
Rati
oof
loca
lex
pen
dit
ure
toto
tal
budget
ary
reven
ue
Fin
anci
al
Sta
tist
ical
Mate
rial
for
Pre
fect
ure
s,C
itie
s,and
Counti
esN
ati
onw
ide
GD
Pp
erca
pit
aL
ogari
thm
of
GD
Pp
erca
pit
ain
aci
tyC
hin
aC
ity
Sta
tist
ical
Yea
rbook
Urb
aniz
ati
on
Rati
oof
urb
an
popula
tion
toth
eto
tal
popula
tion
Chin
aC
ity
Sta
tist
ical
Yea
rbook
Cit
yD
um
my
Dum
my
vari
able
for
the
19
citi
esat
pro
vin
cial
or
dep
uty
-pro
vin
cial
level
Touri
stra
tio
The
rati
oof
rece
ived
touri
sts
toth
eto
tal
popula
tion
Chin
aT
ouri
smSta
tist
ical
Yea
rbook
Sec
reta
ryse
cond
term
indic
ato
rD
um
my
vari
able
indic
ati
ng
ifth
ese
cret
ary
serv
eshis
seco
nd
term
,co
ded
as
1if
tenure
isgre
ate
rth
an
5and
0oth
erw
ise
Com
piled
by
the
auth
or
from
vari
ous
sourc
es
Sec
reta
ryage
of
2005
The
age
of
secr
etari
esas
of
2005
Com
piled
by
the
auth
or
from
vari
ous
sourc
esSec
reta
ryed
uca
tion
Educa
tion
level
of
secr
etari
es,
coded
as
1fo
rdoct
or,
2fo
rM
ast
er,
3fo
rco
lleg
e,and
4fo
roth
ers
The
Worl
dB
ank
surv
ey
Sec
reta
rylo
cal
pro
moti
on
Dum
my
vari
able
indic
ati
ng
ifth
ese
cret
ary
was
loca
lly
pro
mote
d,
coded
as
1fo
rlo
cal
pro
moti
on
and
0oth
erw
ise.
The
Worl
dB
ank
surv
ey
38
Works Cited
Official Documents
CCP Central Committee. Zhonggong zhongyang guanyu jianli lao ganbu tuixiu zhidu de jueding(Decision on Institutionalization of Retirement of Aging Cadres). February 20 1982.
CCP Central Committee. Dangzheng lingdao ganbu xuanba renyong gongzuo zanxing tiaoli(Interim Regulations on Selection and Appointment of Leading Party and Government Cadres).February 9 1995.
CCP Central Committee. Dangzheng lingdao ganbu xuanba renyong gongzuo tiaoli (Regulationson Selection and Appointment of Leading Party and Government Cadres). July 9 2002.
General Office of the CCP Central Committee. Dangzheng lingo ganbu zhiwu renqi zanxing guiding(Interim Regulations on the Tenure of Leading Party and Government Cadres). August 7 2006.
General Office of the State Council. Guanyu shenru kaizhan tudi shichang zhili zhengdun yangetudi guanli de jinji tongzhi (Urgent Notification on Regulating Land Market and StrengtheningLand Management). April 29 2004.
State Bureau of Land Administration and State Planning Commission. Dongjie fei nongye jianshexiangmu zhanyong gengdi guiding (Regulations on Freezing Non-agricultural ConstructionProject Occupying Arable Land). May 20 1997.
State Council. Chengzhen guoyou tudi shiyongquan churang he zhuanrang zanxing tiaoli (InterimRegulations on Transfers of Urban State-owned Land Use Rights). 1990.
State Council. Guowuyuan guanyu jiaqiang tudi tiaokong youguan went de tongzhi (Circular ofthe State Council on Intensifying Land Control). August 31 2006.
State Council. Guojia gongwuyuan zanxing tiaoli (Interim Regulations on the State Civil Servant).August 14 1993.
State Council. Tudi guanlifa shish tiaoli (Regulations on the Implementation of the LandAdministration Law). December 27 1998.
Ministry of Land and Resources. Zhaobiao paimai guapai churang guoyou jianshe yongdishiyongquan guiding (Provisions on the Assignment of State-owned Construction Land UsingRights through Bid Invitation, Auction, and Quotation). September 28 2007.
National People’s Congress. Yusuan fa (Budget Law). 1994.
National People’s Congress. Tudi guanli fa (Land Administration Law). 2004.
National People’s Congress. Xianfa (the Constitution). 2004.
Articles and Books in Chinese
China City Statistical Yearbook, various years. Beijing: China Statistical Press.
China Fiscal Yearbook, various years. Beijing: Ministry of Finance.
39
China Land and Resources Statistical Yearbook, various years. Beijing: Ministry of Land andResources.
China Rural Statistical Yearbook, various years. Beijing: China Statistical Press.
China Tax Yearbook, various years. Beijing: China Tax Press.
China Tourism Statistical Yearbook, 2006. Beijing: National Tourism Administration.
Financial Statistical Material for Prefectures, Cities, and Counties Nationwide, 2005. Beijing:Ministry of Finance. 2006.
Hu Yueming, Feixiang Chen, Jinggang Li, and Lu Wang. 2009. guangzhou shi tudi liyong zhanlueyanjiu (Analysis of Land Use in Guangzhou). Beijing: China Scientific Technology Press.
Jiang, Xingsan, Shouying Liu, and Qing Li. 2007. “tudi zhidu gaige yu guomin jingji chengzhang(Land Reform and Economic Growth).” guanli shijie (Management World) 9: 1-9.
Zhou, Feizhou. 2006. “fenshuizhi shinian (Ten Years after the Tax Sharing System).” Zhongguoshehui kexue (China Social Science) 6: 100-115.
Articles and Books in English
Bates, Robert H. 1981. Markets and States in Tropical Africa: The Political Basis of AgriculturalPolicies. Berkeley: University of California Press.
Burns, John. 1994. “Strengthening CCP Control of Leadership Selection: The 1990 Nomenklatura.”The China Quarterly 138:458–491.
China Daily. 2010. Rural Land Disputes Lead Unrest.
Clague, Christopher, Philip Keefer, Stephen Knack and Mancur Olson. 1996. “Property andContract Rights in Autocracies and Democracies.” Journal of Economic Growth 1(2):243–276.
Easter, Gerald M. 2002. “Politics of Revenue Extraction in Post-communist States: Poland andRussia Compared.” Political Theory 30(4):599 –627.
Edin, Maria. 2003. “State Capacity and Local Agent Control in China: CCP Cadre Managementfrom a Township Perspective.” The China Quarterly 173:35–52.
Frye, Timothy and Andrei Shleifer. 1997. “The Invisible Hand and the Grabbing Hand.” AmericanEconomic Review 87(2):354–358.
Gehlbach, Scott. 2008. Representation through Taxation: Revenue, Politics, and Development inPostcommunist States. Cambridge: Cambridge University Press.
Guo, Gang. 2009. “China’s Local Political Budget Cycles.” American Journal of Political Science53(3):621–632.
Hsing, You-tien. 2010. The Great Urban Transformation: Politics of Land and Property in China.Oxford: Oxford University Press.
Huang, Yasheng. 2008. Capitalism with Chinese Characteristics: Entrepreneurship and the State.Cambridge: Cambridge University Press.
40
Huang, Yiping and Kunyu Tao. 2010. Causes and Remedies of China’s External Imbalances. ChinaCenter for Economic Research Working Paper No. E2010002, Peking University, Beijing.
Jin, Hehui, Yingyi Qian and Barry R. Weingast. 2005. “Regional Decentralization and FiscalIncentives: Federalism, Chinese Style.” Journal of Public Economics 89:1719–1742.
Kornai, Janos. 1992. The Socialist System: The Political Economy of Communism. Princeton,New Jersey: Princeton University Press.
Lam, Tao-chiu and Hon S. Chan. 1996. “Reforming China’s Cadre Management System.” AsianSurvey 36(4):772–786.
Landry, Pierre F. 2003. “The Political Management of Mayors in Post-Deng China.” TheCopenhagen Journal of Asian Studies 17:31–58.
Landry, Pierre F. 2008. Decentralized Authoritarianism in China: The Communist Party’s Controlof Local Elites in the Post-Mao Era. Cambridge: Cambridge University Press.
Levi, Margaret. 1988. Of Rule and Revenue. Berkeley: University of California Press.
Li, Hongbin and Li-An Zhou. 2005. “Political Turnover and Economic Performance: The IncentiveRole of Personnel Control in China.” Journal of Public Economics 89:1743–1762.
Lin, George C. S. 2009. Developing China: Land, Politics and Social Conditions. New York:Routledge.
Lorentzen, Peter L., Pierre F. Landry and John K. Yasuda. 2011. Transparent Authoritarianism?:An Analysis of Political and Economic Barriers to Greater Government Transparency in China.
Man, Joyce Yanyun and Yu-Hung Hong. 2011. China’s Local Public Finance in Transition.Cambridge: Lincoln Institute of Land Policy.
Manion, Melanie. 1985. “The Cadre Management System, Post-Mao: The Appointment,Promotion, Transfer and Removal of Party and State Leaders.” The China Quarterly(102):203–233.
Manion, Melanie. 1993. Retirement of Revolutionaries in China: Public Policies, Social Norms,Private Interests. Princeton, New Jersey: Princeton University Press.
Montinola, Gabriella, Yingyi Qian and Barry R. Weingast. 1996. “Federalism, Chinese Style: ThePolitical Basis for Economic Success in China.” World Politics 48(01):50–81.
Murphy, Kevin M., Andrei Shleifer and Robert W. Vishny. 1991. “The Allocation of Talent:Implications for Growth.” The Quarterly Nournal of Economics 106(2):503–530.
Naughton, Barry. 1992. “Implications of the State Monopoly over Industry and Its Relaxation.”Modern China 18(1):14–41.
North, Douglass C. 1981. Structure and Change in Economic History. New York: W. W. Norton& Company.
Ofer, Gur. 1987. “Soviet Economic Growth: 1928-1985.” Journal of Economic Literature25(4):1767–1833.
41
Oi, Jean C. 1992. “Fiscal Reform and the Economic Foundations of Local State Corporatism inChina.” World Politics 45(1):99–126.
Oi, Jean C. 1999. Rural China Takes Off: Institutional Foundations of Economic Reform. Berkeley:University of California Press.
Oi, Jean C. and Shukai Zhao. 2007. Fiscal Crisis in China’s Township: Causes and Consequences. InGrassroots Political Reform in Contemporary China, ed. Elizabeth J. Perry and Merle Goldman.Cambridge: Harvard University Press pp. 75–96.
Olson, Mancur. 1993. “Dictatorship, Democracy, and Development.” The American PoliticalScience Review 87(3):567–576.
Pollack, Sheldon D. 2009. War, Revenue, and State Building. Ithaca: Cornell University Press.
Roland, Gerard. 2000. Transition and Economics: Politics, Markets, and Firms. Cambridge,Massachusetts: The MIT Press.
Sheng, Yumin. 2011. Economic Openness and Territorial Politics in China. Cambridge: CambridgeUniversity Press.
Shih, Victor, Christopher Adolph and Mingxing Liu. 2011. Getting Ahead in the Communist Party:Explaining the Advancement of Central Committee Members in China.
Shih, Victor, Luke Qi Zhang and Mingxing Liu. 2010. When the Autocrat Gives: Determinants ofFiscal Transfers in China.
Shleifer, Andrei. 1997. “Government in transition.” European Economic Review 41:385–410.
Shleifer, Andrei and Robert W. Vishny. 1998. The Grabbing Hand. In The Grabbing Hand:Government Pathologies and Their Cures, ed. Andrei Shleifer and Robert W. Vishny. Cambridge,Massachusetts: Harvard University Press.
Tao, Ran, Fubing Su, Mingxing Liu and Guangzhong Cao. 2010. “Land Leasing and Local PublicFinance in China’s Regional Development: Evidence from Prefecture-level Cities.” Urban Studies47(10):2217 –2236.
Tsui, Kai-yuen and Youqiang Wang. 2004. “Between Separate Stoves and A Single Menu: FiscalDecentralization in China.” The China Quarterly 177:71–90.
Whiting, Susan H. 2000. Power and Wealth in Rural China: The Political Economy of InstitutionalChange. Cambridge: Cambridge University Press.
Whiting, Susan H. 2004. The Cadre Evaluation System at the Grassroots. In Holding ChinaTogether: Diversity and National Integration in the Post-Deng Era, ed. Barry J. Naughton andDali L. Yang. Cambridge: Cambridge University Press.
Whiting, Susan H. 2011. Fiscal Reform and Land Public Finance: Zouping County in NationalContext. In China’s Local Public Finance in Transition, ed. Joyce Yanyun Man and Yu-HungHong. Lincoln Institute of Land Policy pp. 125–144.
Wong, Christine. 2009. “Rebuilding Government for the 21st Century: Can China IncrementallyReform the Public Sector?” The China Quarterly 200:929–952.
42
World Bank. 2002. China National Development and Subnational Finance. Washington, DC: WorldBank.
World Bank. 2006. China Governance, Investment Climate, and Harmonious Society: CompetitiveEnhancement for 120 Cities in China.
Wright, Joseph. 2008. “To Invest or Insure? How Authoritarian Time Horizons Impact ForeignAid Effectiveness.” Comparative Political Studies 41(7):971 –1000.
Yang, Dali L. 2004. Remaking the Chinese Leviathan: Market Transition and the Politics ofGovernance in China. Stanford: Stanford University Press.
43